Studies Of Mortality Prediction Models And Individualized Treatments Based On Investigation Of Pediatric Septic Shock In Jilin Province

Sepsis is life threatening organ dysfunction caused by a dysregulated host response to infection.Septic shock,the most severe subtype of sepsis,is characterized by rapid progression and high mortality.To reduce mortality from sepsis,the World Health Assembly has launched a series of initiatives,including:developing national(or regional)policies to improve sepsis care,training health care workers,promoting scientific research about innovative means of prevention,diagnosis,treatment and prediction of sepsis.Therefore,the objective of this study was to reduce mortality from pediatric septic shock,and the following studies were conducted from three aspects:the distribution of pediatric critical medical resources,physicians’ cognition of septic shock and the characteristics of septic shock.Chapter 1 Investigation of Pediatric Septic Shock in Jilin ProvinceObjective:There are few multicenter studies on pediatric septic shock in China.This study investigated the data of pediatric septic shock in secondary and tertiary hospitals in Jilin Province.It was the first domestic investigation of pediatric septic shock covering secondary and tertiary hospitals in Jilin province,and improved the data of pediatric septic shock in Jilin Province.Methods:Retrospective observational study.Relying on the organizational network of Pediatric Quality Control Center of Jilin Province,the data of patients admitted to the pediatrics department of secondary and tertiary hospitals and diagnosed with septic shock from January 2018 to December 2022 were retrospectively collected.Results:(1)General data:242 cases of pediatric septic shock were admitted in 8 hospitals in 7 cities of Jilin Province,82.6%of which came from the First Hospital of Jilin University.The most common primary site of infection was the respiratory system(33.1%).(2)About bundles:the completion rates of patients in Changchun in initiating fluid resuscitation,lactic acid monitoring,blood culture collection,blood culture collection before antibiotics,and application of vasoactive drugs were better than those not in Changchun.(3)Outcomes:the fatality rate was 36%.29.9%died within 24 hours and 57.5%within 3 days after admission.Since 2018,the case fatality rate has shown a gradually decreasing trend.Conclusion:The fatality rate of pediatric septic shock was high.The source of the cases was highly concentrated in tertiary hospitals in Changchun.There were regional differences in the completion rates of bundles for pediatric septic shock.Chapter 2 Investigation on the Knowledge of Pediatricians to Septic Shock and the Related Pediatric Critical Medical Resources in Jilin ProvinceObjective:Since the mortality of pediatric septic shock is also determined by the physician’s knowledge about pediatric septic shock and the intensive care medical resources,we investigated both.This will lay the foundation for developing systematic training programs for pediatricians,improving the medical quality of pediatric septic shock and reducing mortality.Methods:A cross-sectional survey.Firstly,pediatricians in the top 25%tertiary hospitals and/or specialized pediatric hospitals in each city of Jilin Province were surveyed in the form of questionnaires on their knowledge for septic shock.Secondly,the pediatric critical care resources for septic shock were investigated in the only 5 hospitals with pediatric intensive care units.Results:Twenty centers participated in this study,and 461 valid questionnaires were collected.The training method of the surveyed pediatricians was mainly self-study(69.4%),only 5.4%of the pediatricians accurately selected the diagnostic criteria,and only 2.4%of the pediatricians accurately selected the treatment bundles.Twenty hospitals were able to detect lactic acid and provide ECG and oxygen monitoring,16(80%)were able to perform deep vein catheterization,and 11(55%)were able to monitor arterial blood pressure and central venous pressure.Of the 5 hospitals with independent pediatric intensive care units in Jilin province,2 were located in Changchun.There are the most pediatric intensive care beds(63.6%),critically ill children(75.6%),professors or associate professors(45%),attending physicians(64.7%)and physicians(75%)in Changchun.In terms of educational background,pediatricians with master’s degree or above were also mainly in Changchun(95.5%)Conclusion:Pediatricians in Jilin Province still have insufficient knowledge of pediatric septic shock and lack systematic training.Some hospitals still lack the equipments related to septic shock.High-quality pediatric intensive care resources are concentrated in Changchun.In the future,it is necessary to strengthen the training of pediatricians about septic shock.Chapter 3 Comparison of Predictive Value of Machine Learning Algorithm for Pediatric Septic Shock Mortality RiskObjective:The mortality of pediatric septic shock is high.Early prediction of the mortality risk of pediatric septic shock is helpful for clinicians to judge the severity of the disease,take active treatments,and improve the adverse outcome of patients.Machine learning has a good performance in predicting the prognosis of diseases.At present,there are few researches on machine learning algorithm to establish mortality prediction model of pediatric septic shock.This study established three mortality risk prediction models based on random forest,support vector machine and logistic regression,and evaluated the prediction efficiency of different models.Methods:In this study,a multi-center pediatric septic shock database was established in Jilin Province.The pediatric septic shock of the First Hospital of Jilin University from January 2012 to December 2021 were used as a model derived set,which was randomly divided into a training set and a test set(7:3)for the establishment and testing of models,including random forest,support vector machine and Logistic regression.Pediatric septic shock data from the First Hospital of Jilin University and Children’s Hospital of Changchun from January 2022 to present were used as the validation set for the three models.Compare the effects of the three models in the test set and validation set.Results:284 pediatric septic shock patients were included in the model derivation group,and 140(49.3%)died during hospitalization.The model derivation group was randomly divided into a training set(n=199)and a test set(n=85)at 7:3.Ten patients(31.3%)in the validation group(n=32)died during hospitalization.No matter the test set or the validation set,the area under the receiver characteristic curve of random forest and support vector machine is higher than that of Logistic regression model.The factors that predicted death in the three models were lactic acid,infection site and underlying diseases.Conclusion:Random forest and support vector machines were superior to traditional Logistic regression models in predicting pediatric septic shock mortality.Chapter 4 Lactic Trajectory-based Latent Class Mixed Model and Individualized Management of Pediatric Septic ShockObjective:Due to the heterogeneity of septic shock,one treatment plan is not appropriate for all.Classifying pediatric septic shock according to certain characteristics and applying individualized treatments may reduce mortality.At present,there are few reports on the classification of pediatric septic shock and the exploration of individualized treatments.In this study,a latent class mixed model was established based on lactic acid to classify pediatric septic shock,and a new idea of individualized treatment was explored.Methods:Single-center,retrospective,observational study.Pediatric septic shock admitted to the pediatric intensive care unit of the First Hospital of Jilin University from January 1,2012 to February 28,2023 were studied.A latent class mixed model was established based on lactic acid within 72 hours after the diagnosis of septic shock.Pediatric septic shock was classified according to the trajectory of lactic acid,and the clinical characteristics and treatment regiments of different subtypes were compared,and the influence of treatment regiments on the mortality within different subtypes was discussed.Results:(1)Pediatric septic shock was divided into three subgroups:the first group(accounted for 4%)was characterized by low lactate level within 72 hours.The fatality rate was 36%and 82%patients with multiple organ dysfunction syndrome.The second group(accounted for 70%)was characterized by initial high lactic acid,but gradually decreased to normal.The fatality rate was 35%and 50.5%patients with multiple organ dysfunction syndrome.The third group(accounted for 26%)was characterized by initial high lactic acid,which increased over time,with fatality rate of 92%and 81%of the patients with multiple organ dysfunction syndrome in the third group.(2)In the first and second groups of septic shock,respiratory failure,thrombocytopenia,and septic shock due to nosocomial infection increased mortality risk(P<0.05).Increased fluid volume(up to 40 ml/kg.h)in the first hour of resuscitation and was collected blood culture before antibiotics was associated with reduced mortality risk.Conclusion:Pediatric septic shock can be divided into different phenotypes according to lactic acid trajectory.Different phenotypes have different clinical characteristics and may require different treatment regiments.